bjutsecurity22
/
mindspore2022
forked from huawei/mindspore2022
0
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

add check for group parameters

This commit is contained in:
guohongzilong 2020-05-26 22:15:23 +08:00
parent c6f309e125
commit 75c1e7f6af
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
mindspore/nn/optim/optimizer.py
View File

@ -112,7 +112,7 @@ class Optimizer(Cell):
learning_rate = self._get_single_lr(learning_rate)
if isinstance(parameters[0], dict):
self.is_group = True
self.params = []
self.group_params = []
self.group_lr = []
self.group_weight_decay = []
self._init_group_params(parameters, learning_rate, weight_decay)
@ -123,7 +123,7 @@ class Optimizer(Cell):
self.learning_rate = Parameter(learning_rate, name="learning_rate")
if self.is_group:
self.parameters = ParameterTuple(self.params)
self.parameters = ParameterTuple(self.group_params)
self.weight_decay = tuple(self.group_weight_decay)
decay_filter = lambda x: x > 0
self.decay_flags = tuple(decay_filter(x) for x in self.weight_decay)
@ -230,7 +230,10 @@ class Optimizer(Cell):
params_store = []
for group_param in parameters:
self.params += group_param['params']
if not group_param['params']:
raise ValueError("Optimizer got an empty parameter list.")
self.group_params += group_param['params']
if 'lr' in group_param.keys():
params_dynamic_lr = isinstance(group_param['lr'], (Iterable, Tensor))